I know little about EEG signals. My project is about the classification problem of sleep time signals (5-classification, wake, NREM1, NREM2, NREM3, REM). I extract features from ECG signals for classification, but the result of experiments is poor. I want to extract features from the EEG signal for testing, but I do not know how to extract features from the EEG.
I use SHHS sleep datasets (https://sleepdata.org/datasets/shhs/). I use the MNE package in the python environment to pre-process the raw EDF signals from the SHHS dataset, for example, a total of 3596250 data in the raw, split into 30-s epochs for 959 epochs, with 3750 data per epoch. As a follow-up, is there any way to extract the features?
To get better results you may convert the signals to 2D images and then apply the CNN model like- VGG-19, ResNet, Inception, etc.
Hello! I suggest using TSFEL (https://github.com/fraunhoferportugal/tsfel) - an open-source Python library optimized for time series that automatically extracts over 60 different features on the statistical, temporal and spectral domains.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series